International Journal of Trend in Scientific Research and Development (IJTSRD)
Volume 5 Issue 2, January-February 2021 Available Online: www.ijtsrd.com e-ISSN: 2456 – 6470
A Survey on Heart Disease Prediction Techniques
G. Niranjana, Dr I. Elizabeth Shanthi
Department of Computer Science, Avinashilingam Institute for Home
Science and Higher Education for Women, Coimbatore, Tamil Nadu, India
ABSTRACT
Heart disease is the main reason for a huge number of deaths in the world
over the last few decades and has evolved as the most life-threatening disease.
The health care industry is found to be rich in information. So, there is a need
to discover hidden patterns and trends in them. For this purpose, data mining
techniques can be applied to extract the knowledge from the large sets of data.
Many researchers, in recent times have been using several machine learning
techniques for predicting the heart related diseases as it can predict the
disease effectively. Even though a machine learning technique proves to be
effective in assisting the decision makers, still there is a scope for developing
an accurate and efficient system to diagnose and predict the heart diseases
thereby helping doctors with ease of work. This paper presents a survey of
various techniques used for predicting heart disease and reviews their
performance.
How to cite this paper: G. Niranjana | Dr I.
Elizabeth Shanthi "A Survey on Heart
Disease Prediction Techniques" Published
in
International
Journal of Trend in
Scientific Research
and
Development
(ijtsrd), ISSN: 24566470, Volume-5 |
Issue-2,
February
IJTSRD38349
2021, pp.68-71, URL:
www.ijtsrd.com/papers/ijtsrd38349.pdf
Copyright © 2021 by author(s) and
International Journal of Trend in Scientific
Research and Development Journal. This
is an Open Access article distributed
under the terms of
the
Creative
Commons Attribution
License
(CC
BY
4.0)
KEYWORDS: Heart disease, data mining, machine learning, prediction
(http://creativecommons.org/licenses/by/4.0)
1. INTRODUCTION
Data mining is the process of examining large databases in
order to discover the hidden patterns and correlations using
statistics, machine learning, artificial intelligence and
database technology. Tremendous amount of data is
generated in medical field and it is important to mine those
data for helping the practitioners in early diagnosis of
disease. Heart disease causes immediate death and claims
more lives each year than compared to all types of cancer or
other major diseases. Heart disease prediction is very
challenging area in the medical field because of several risk
factors such as high blood pressure, high cholesterol,
uncontrolled diabetes, abnormal pulse rate, obesity etc.
Highly skilled and experienced physicians are required to
diagnose the heart disease [1]. However, the death rate can
be drastically reduced if the disease is detected at the early
stages and also by adopting preventive measures. So,
developing a heart disease prediction system is
indispensable thereby preventing the patient’s death
through early diagnosis. The main objective of this paper is
to do a survey on the previous research work in heart
disease prediction and analyzing the techniques used.
The organization of the paper is as follows. Section 2 tells
about the heart disease. Section 3 explains about the data
mining algorithms for heart disease prediction. Section 4
deals with the literature review. Section 5 shows the
observation and section 6 concludes the paper.
2. HEART DISEASE
Heart disease occurs when plaque develops in the arteries
and blood vessels that lead to the heart. This plaque blocks
@ IJTSRD
|
Unique Paper ID – IJTSRD38349
|
oxygen and important nutrients from reaching your heart.
The types of heart disease are listed below:
1. Congenital heart disease
2. Arrhythmia
3. Coronary artery disease
4. Dilated cardiomyopathy
5. Myocardial infarction
6. Mitral valve prolapse
7. Pulmonary stenosis
8. Hypertrophic cardiomyopathy
3. DATA MINING TECHNIQUES FOR PREDICTION
3.1. DECISION TREE
Decision Tree is a type of Supervised Machine Learning
algorithm, where it identifies several ways to split the data
continuously based on certain parameter. The tree consists
of two entities, namely decision node and leaf node. The
decision nodes are where the data is split and the leaf node
are the decisions or the final outcomes. Decision tree
performs both classification and regression tasks. The tree
model where the target variable takes a set of discrete values
are called classification trees and when the target variable
takes continuous values are called as regression trees.
Decision tree is widely used in data mining, statistics and
machine learning.
3.2. NAÏVE BAYES
A Naive Bayes classifier is a probabilistic machine learning
algorithm based on Bayes theorem for binary and multi-class
classification problems. A Naive Bayesian classifier is easy to
build, with independent assumptions between predictors.
Volume – 5 | Issue – 2
|
January-February 2021
Page 68
International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
This assumption is most unlikely in real data. In Naive Bayes
every pair of features being classified is independent of each
other. It outperforms many other sophisticated classification
models. Bayes’ theorem is stated as follows
P(A/B) = P(B/A).P(A)/P(B)
for classification after deducting the feature with lowest
information gain. Now, the accuracy is 89.56% in training
dataset and 80.99% in validation dataset. The result shows
that feature selection helps increase computational
efficiency while improving classification accuracy.
3.3. Random Forest
Random forest is a supervised learning algorithm which
operates by constructing multiple decision trees for both
classification as well as regression. The random forest
algorithm creates decision trees based on data samples and
then predicts the output for each of them and voting is done
finally to select the best solution or output. The main
advantage of using random forest is that it reduces the overfitting and provides high accuracy.
Nikhil Gawande et al. [4] (2017) proposed a system to
classify heart disease using convolutional neural network.
The system uses a CNN model to classify ECG signals. As it
can be able to classify heart beat in different manner, there is
no need for feature extraction. ECG signals have been taken
from the MIT-BIH database. ECG signal is given as input to
the system. Total 340 samples are trained and tested using
CNN and the results are described in the confusion matrix.
Even long ECG records can be classified in accurate manner
and the distinct patient’s records can be treated once the
CNN is trained. The accuracy obtained is 99.46%.
3.4. SUPPORT VECTOR MACHINE
A Support vector machine is a discriminative classifier that
finds a hyper plane in n-dimensional space to distinctly
classify the data points. Given a labelled training data, the
algorithm outputs an optimal separating hyper plane which
divides the data or classes. Support vectors are the data
points or class closer to the hyper plane. SVM is a supervised
learning method used for both classification and regression
tasks. It is widely used algorithm since it produces high
accuracy with less computation power.
3.5. ARTIFICIAL NEURAL NETWORK
Neural network is a machine learning algorithm that is
designed based on the human brain. Artificial neural
network is an information processing technique that works
like the way human brain process the information. ANN
contains 3 layers input layer, hidden layer and the output
layer that are interconnected by nodes which contains an
activation function. Neural network is greatly used in data
mining field for classifying large datasets. It enhances the
data analysis technology.
4. LITERATURE REVIEW
In this section a detailed description of the previous work
that has gone into the research related to the data mining
technique for heart disease prediction are explained.
Purushottam et al. [2] (2015) designed a system that can
efficiently discover the rules to predict the risk level of heart
disease based on the given parameter. The rules are
prioritized based on the user’s requirements. The system
uses the classification model by covering rules as C4.5Rules.
WEKA tool is used for dataset analysis and Knowledge
Extraction based on Evolutionary Learning (KEEL) tool to
find out the classification decision rules. The dataset is taken
from the Cleveland Clinic Foundation. It contains total 76
raw attributes, out of which only 14 of them are taken. This
dataset contains several parameters like ECR, cholesterol,
chest pain, fasting sugar, MHR and many more. The
classification result of the decision tree is 87.4%.
AnchanaKhemphila et al. [3] (2011) proposed a classification
approach using Multi-Layer Perceptron with Back
Propagation learning algorithm and a feature selection
algorithm to diagnose heart disease. Information Gain
concept is used for feature selection. First the model uses the
ANN with no information gain-based feature selection
function; the accuracy in training dataset is 88.46% and
80.17% in the validation dataset. Further, the ANN is used
@ IJTSRD
|
Unique Paper ID – IJTSRD38349
|
V Krishnaiah et al. [5] (2014) introduces a fuzzy
classification technique to diagnose the heart disease. The
main objective of the research work is to predict the heart
disease patients with more accuracy and to remove the
uncertainty in unstructured data. The data source is
collected from Cleveland Heart Disease database and Stalog
Heart disease database. Cleveland database consist of 303
records and Stalog database consists of 270 records and
remaining records are collected from different hospitals in
Hyderabad.
M.A. Jabbar et al. [6] (2016) proposed a novel classification
model Hidden Naïve Bayes classifier for heart disease
prediction. The heart dataset is downloaded from the UCI
repository. Heart stalog dataset contains 14 attributes and
270 instances. Hidden Naïve Bayes evaluation is performed
using 10-fold cross validation. WEKA tool is used for hidden
naïve bayes classification. The proposed approach performs
pre-processing using discretization and IQR filters to
improve the efficiency of Hidden Naïve Bayes. The
performance results show that the HNB classifier model
recorded 100% accuracy compared with NB classification
model.
Jayshril S. Sonawane et al. [7] (2014) proposed a system to
predict heart disease using multilayer perceptron
architecture of neural network. The dataset is taken from the
Cleveland heart disease database. The dataset containing 13
clinical attributes fed as input to the neural network and
back propagation algorithm is used to train the data.
Compared to the decision support system for predicting
heart disease using multilayer perceptron and back
propagation algorithm, the proposed system achieves
highest accuracy of 98.58% for 20 neurons.
TanmayKasbe et al. [8] (2017) proposed fuzzy expert system
for heart disease diagnosis. The database has been taken
from the UCI repository and this dataset consist of 4
databases that are taken from V.A. Medical center, Long
Beach, Cleveland clinic foundation, Hungarian institute of
cardiology, Budapest and University hospital, Zurich,
Switzerland. A total of 76 input attributes and 1 output
attribute are in the dataset, out of this the proposed system
uses 10 important input attributes and 1 output attribute.
MATLAB software is used for developing fuzzy system. The
fuzzy system consists of 3 steps i.e. fuzzification, rule base
and defuzzification. The proposed fuzzy expert system
Volume – 5 | Issue – 2
|
January-February 2021
Page 69
International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
achieves 93.33% accuracy and better performance
compared to previous work on same domain.
dataset is small in size. There is no clear information about
the accuracy level.
Aakashchauhan et al. [9] (2018) proposed a new rule to
predict the coronary heart disease using evolutionary rule
learning. Computational intelligence is used to discover the
relationship between disease and patient. The dataset is
taken from the Cleveland Heart Disease database. KEEL is a
java-based tool used for the simulation of evolutionary
learning. Classification model was developed by association
rule and the results are evaluated. The system will help
doctor to explore their data and predicts the coronary
disease accurately.
M. AkhilJabbar et al, [14] (2013) proposed a neural network
model to classify the heart disease using artificial neural
network and feature subset selection. The feature subset
selection is a method that is used to reduce the
dimensionality of the input data. By reducing the number of
attributes, the number of diagnosis tests which are needed
by doctors from patient are also reduced. The dataset is
taken from Andhra Pradesh hospital and results show that
accuracy is enhanced over the out-dated classification
techniques. The results also show that this system is faster
and precise.
Senthilkumarmohan et al. [10] (2019) proposed a novel
method to improve the prediction accuracy of cardiovascular
disease using hybrid machine learning techniques. The
prediction model is designed with different combinations of
features and several known machine learning classification
techniques. Cleveland Heart Disease dataset is taken from
the UCI repository. After feature selection 13 attributes are
considered for further classification. R studio rattle is used
for performing classification and the performance are
evaluated. The prediction model uses hybrid random forest
with a linear model and recorded 88.7% accuracy.
Sinkonnayak et al. [11] (2019) proposed a method to predict
the heart disease by mining frequent items and classification
techniques. The dataset is taken from UCI repository and
pre-processing is done. The frequent item mining is used for
filtering the attributes and then the variant classification
techniques like Decision tree, Naïve Bayes, Support Vector
Machine and KNN classification methods are used for
predicting the heart disease at an early stage. R analytical
tool is used for implementation. Out of these diverse data
mining techniques Naïve Bayes achieves 88.67% accuracy in
predicting the heart disease with attribute filtration. The
performance is evaluated using ROC curve.
RahmaAtallah et al. [12] (2019) proposed a majority voting
ensemble method to predict the heart disease in humans.
Here the model classifies the patient based on the majority
vote of diverse machine learning models in order to provide
more accurate results. The dataset is taken from UCI
repository and 14 predominant attributes are considered. In
pre-processing Min-Max normalization is done. In order to
analyze the data, a correlation value was calculated between
each attribute and the target variable. It can be noted that
the highest correlated features with the target attribute were
Cp, Thalach, Oldpeak and Exang. For testing 4 types of
classifier models are used namely Stochastic Gradient
Descent (88%), KNN (87%), Logistic regression (87%) and
Random forest (87%). These 4 models are combined in an
ensemble model where the classification is done based on
hard voting and finally the model achieves 90% accuracy.
Haritajagad and Jehankandawalla et al, [13] (2015) proposed
a model to detect coronary artery disease using different
data mining algorithms namely Decision Tree, Naïve Bayes
and Neural Network. The parameters like patient age, sex,
blood pressure etc. is taken during check up to evaluate the
performance of these algorithms. Naïve Bayes proved to be
the fastest among three algorithms. Decision tree algorithm
reliability depends on the input data and it is difficult to deal
with large datasets. Neural network is basically used when
@ IJTSRD
|
Unique Paper ID – IJTSRD38349
|
Muhammad saqlain and wahidhussain et al, [15] (2016)
proposed a multi nominal Naïve Bayes algorithm to detect
the heart failure. The data are collected from Armed Forces
Institute of Cardiology (AFIC), Pakistan in the form of
medical records. It uses 30 variables. The proposed
algorithm is compared with different classification
algorithms like Logistic Regression, Neural Network, SVM,
Random Forest and Decision Tree. The performance of
Navies Bayes algorithm is measured in terms of Precision,
Accuracy, Recall and Area Under the Curve (AUC). Naïve
Bayes achieved highest accuracy of 86.7% and Area under
the Curve (AUC) is 92.4% respectively.
5. OBSERVATION
From the literature review, it is observed that the dataset
was taken from UCI repository containing Cleveland Heart
Disease database and Stalog Heart disease database. The
Naïve Bayes, Decision tree, Random forest and neural
network algorithms are predominantly used data mining
techniques for predicting the heart disease with highest
accuracy. But the authors should concentrate on minimizing
the time utilization. Most of the research works have used
existing machine learning techniques or combined two or
more existing techniques to improve the prediction
performance. Rather than taking vote or combining the
existing algorithms the researchers should propose new
algorithms for prediction. The researchers should explore
more deep learning concepts in order toimprove the
prediction performance. Since the heart disease is very
dangerous a fast and reliable system need to be developed.
6. CONCLUSION
The main motive for conducting this survey is to comprehend
the work of different authors and also to analyze how
accurately we can predict the heart disease. These authors
have proposed different data mining algorithms but there are
certain limitations like neural networks performs well only
with structured dataset [16]. MATLAB, Python and WEKA
tool are widely used technologies for implementing these
algorithms. However different algorithms generate different
accuracy that purely depends on the size of the dataset,
number of attributes selected [17] and tools used for
implementation. The future works should concentrate on
proposing new algorithms to achieve better performance and
to minimize the time utilization.
REFERENCES
[1] M. A. Jabbar et.al” Prediction of heart disease using
Random forest and Feature subset selection “,AISC
SPRINGER, vol 424,pp187-196(2015)
Volume – 5 | Issue – 2
|
January-February 2021
Page 70
International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
[2]
Purushottam, K. Saxena and R. Sharma, "Efficient heart
disease prediction system using decision tree,"
International
Conference
on
Computing,
Communication & Automation, Noida, 2015, pp. 72-77.
Learning Techniques," in IEEE Access, vol. 7, pp.
81542-81554, 2019.
[11]
S. Nayak, M. K. Gourisaria, M. Pandey and S. S.
Rautaray, "Prediction of Heart Disease by Mining
Frequent Items and Classification Techniques," 2019
International Conference on Intelligent Computing and
Control Systems (ICCS), Madurai, India, 2019, pp. 607611, doi: 10.1109/ICCS45141.2019.9065805.
[12]
R. Atallah and A. Al-Mousa, "Heart Disease Detection
Using Machine Learning Majority Voting Ensemble
Method," 2019 2nd International Conference on new
Trends in Computing Sciences (ICTCS), Amman,
Jordan,
2019,
pp.
1-6,
doi:
10.1109/ICTCS.2019.8923053.
[3]
Khemphila and V. Boonjing, "Heart Disease
Classification Using Neural Network and Feature
Selection," 2011 21st International Conference on
Systems Engineering, Las Vegas, NV, 2011, pp. 406409.
[4]
N. Gawande and A. Barhatte, "Heart diseases
classification using convolutional neural network,"
2017 2nd International Conference on Communication
and Electronics Systems (ICCES), Coimbatore, 2017,
pp. 17-20.
[5]
V. Krishnaiah, M. Srinivas, G. Narsimha and N. S.
Chandra, "Diagnosis of heart disease patients using
fuzzy classification technique," International
Conference on Computing and Communication
Technologies, Hyderabad, 2014, pp. 1-7.
[13]
HaritaJagad and JehanKandawalla , “ Detection of
Coronary Artery Diseases using Data Mining
Techniques”, International Journal on Recent And
Innovation Trends in Computing And Communication
ISSN:2321-8169 Vol.3 issue 11,pp.60906092,2015.
[6]
M. A. Jabbar and S. Samreen, "Heart disease prediction
system based on hidden naïve bayes classifier," 2016
International Conference on Circuits, Controls,
Communications and Computing (I4C), Bangalore,
2016, pp. 1-5.
[14]
Jabbar, M. Akhil, B. L. Deekshatulu, and Priti Chandra.
"Classification of heart disease using artificial neural
network and feature subset selection."Global Journal
of Computer Science and Technology Neural &
Artificial Intelligence 13, no. 3 (2013).
[7]
J. S. Sonawane and D. R. Patil, "Prediction of heart
disease using multilayer perceptron neural network,"
International
Conference
on
Information
Communication
and
Embedded
Systems
(ICICES2014), Chennai, 2014, pp. 1-6.
[15]
[8]
T. Kasbe and R. S. Pippal, "Design of heart disease
diagnosis system using fuzzy logic," 2017
International Conference on Energy, Communication,
Data Analytics and Soft Computing (ICECDS), Chennai,
2017, pp. 3183-3187.
Muhammad Saqlain, Wahid Hussain, Nazar A. Saqib et
al, “identification of heart failure by using
unstructured data of cardiac patients”, 45th
International Conference on Parallel Processing
Workshops,
DOI
10.1109/ICPPW.2016.66,
pp.426431,2016.
[16]
B. Gnaneswar and M. R. E. Jebarani, "A review on
prediction and diagnosis of heart failure," 2017
International Conference on Innovations in
Information, Embedded and Communication Systems
(ICIIECS), Coimbatore, 2017, pp. 1-3.
[17]
A. L. Rane, "A survey on Intelligent Data Mining
Techniques used in Heart Disease Prediction," 2018
3rd International Conference on Computational
Systems and Information Technology for Sustainable
Solutions (CSITSS), Bengaluru, India, 2018, pp. 208213.
[9]
[10]
A. Chauhan, A. Jain, P. Sharma and V. Deep, "Heart
Disease Prediction using Evolutionary Rule Learning,"
2018 4th International Conference on Computational
Intelligence & Communication Technology (CICT),
Ghaziabad, 2018, pp. 1-4.
S. Mohan, C. Thirumalai and G. Srivastava, "Effective
Heart Disease Prediction Using Hybrid Machine
@ IJTSRD
|
Unique Paper ID – IJTSRD38349
|
Volume – 5 | Issue – 2
|
January-February 2021
Page 71